// @ts-ignore
import { containerBootstrap } from "@nlpjs/core";
// @ts-ignore
import { Nlp } from "@nlpjs/nlp";
// @ts-ignore
import { NluManager, NluNeural } from "@nlpjs/nlu";
// @ts-ignore
import { LangEn } from "@nlpjs/lang-en-min";
// @ts-ignore
import { LangZh } from "@nlpjs/lang-zh";
import fs from "node:fs";
import * as fflate from 'fflate';

let zh: TrainData = {};
let en: TrainData = {};

type TrainData = {
    [key: string]: string[];
};

export async function trainIntentionModel() {
    try {
        const dataZH = fs.readFileSync("./lib/nlp/data/zh.json", "utf8");
        const dataEN = fs.readFileSync("./lib/nlp/data/en.json", "utf8");
        zh = JSON.parse(dataZH);
        en = JSON.parse(dataEN);
    } catch (err) {
        console.error(err);
    }

    const container = await containerBootstrap();
    container.use(Nlp);
    container.use(LangEn);
    container.use(LangZh);
    container.use(NluNeural);
    const manager = new NluManager({
        container,
        locales: ["en", "zh"],
        nlu: {
            useNoneFeature: true
        }
    });
    // Adds the utterances and intents for the NLP

    for (const key in zh) {
        for (const value of zh[key]) {
            manager.add("zh", value, key);
        }
    }

    for (const key in en) {
        for (const value of en[key]) {
            manager.add("en", value, key);
        }
    }

    await manager.train();

    // let actual = await manager.process("en", "base64 decode bilibili");
    // console.log(actual);
    // let actualZH = await manager.process("zh", "去除百分号");
    // console.log(actualZH);

    const resultModel = manager.toJSON();

    const buf = fflate.strToU8(JSON.stringify(resultModel));

    const gzipped = fflate.gzipSync(buf, {
        filename: 'model.json',
        mtime: new Date().getTime()
    });
    
    fs.writeFileSync("./public/model", Buffer.from(gzipped)); 
}

trainIntentionModel();